邮箱：econometrics666@sina.cn所有计量经济圈方法论丛的code程序, 宏微观数据库和各种软件都放在社群里.欢迎到计量经济圈社群交流访问.
之前，引荐过“中国学界F4发表AER一篇! 知识青年上山下乡与农村教育问题！”。今天，我们把该文原始数据和运行程序分享给各位学者，有需要的可以根据程序说明运行出结果。
Yi Chen，Ziying Fan，Xiaomin Gu，Li-An Zhou. Arrival of Young Talent: The Send-Down Movement and Rural Education in China, American Economic Review，2020.
This paper estimates the effects of the send-down movement during the Cultural Revolution— when about 16 million urban youth were mandated to resettle in the countryside— on rural education. Using a county-level dataset compiled from local gazetteers and population censuses, we show that greater exposure to the sent-down youths significantly increased rural children's educational achievement. This positive effect diminished after the urban youth left the countryside in the late 1970s but never disappeared. Rural children who interacted with the sent-down youths were also more likely to pursue more-skilled occupations, marry later, and have smaller families than those who did not.
注：为更好地浏览以下code，建议使用电脑查看。
下面展示了1-Table_Census_1990这个程序里的具体code（也是主要结果），其他的可以在文后下载。
******************************************************************************This do-file carries out the analysis using the 1990 census.Table 2.txt (Summary Statistics)Table 3.txt Columns (1)--(7) (The Effect of SDYs on the Educational Attainment of Rural Children)Table 4.txt (Heterogeneous Effect of SDYs)Table 6.txt (Addressing Various Confounding Factors)Table 8.txt Columns (1)--(7) (The Lasting Effect of SDYs on Outcomes other than Education)Figure 3.txt, census 1990 ***********************************************Compute speed of school construction program***********************************************use "$path1B\county_year_data.dta", clearforeach s in secondary primary {bysort countyid: egen Min_year = min(year) if inrange(year,1964,1966)&!missing(school_`s')bysort countyid: egen Max_year = max(year) if inrange(year,1975,1977)&!missing(school_`s')bysort countyid: egen min_year = mean(Min_year)bysort countyid: egen max_year = mean(Max_year)gen Min_school = school_`s' if year == min_yeargen Max_school = school_`s' if year == max_yearbysort countyid: egen min_school = mean(Min_school)bysort countyid: egen max_school = mean(Max_school)gen `s'_speed = (max_school-min_school)/(max_year-min_year)drop Min_* Max_* min_* max_*drop if secondary_speed ==. | primary_speed == .keep countyid secondary_speed primary_speedsave "$path1A\rural_school_expansion.dta", replace********************************************************************************* Step 1: Data Preparation and Summary Statistics *********************************************************************************use "$path1B\census_1990_clean.dta", cleargen treat = inrange(year_birth,1956,1969) if inrange(year_birth,1946,1969)global var_abs_cohort "region1990 prov#year_birth c.primary_base#year_birth c.junior_base#year_birth"global var_abs_cohort2 "region1990 prov#year_birth c.primary_base_older#year_birth c.junior_base_older#year_birth"************************************* Generate County Characteristics *************************************merge m:1 countyid using "$path1A\rural_school_expansion.dta", nogenerate keep(1 3)gen speed_primary_density = primary_speed/(pop1964/1000)gen speed_secondary_density = secondary_speed/(pop1964/1000)*intensity of the Great Faminegen famine = inrange(year_birth,1959,1961) if rural == 1gen nonfamine = inrange(year_birth,1955,1957) if rural == 1bysort region1990: egen sum_famine = sum(famine)bysort region1990: egen sum_nonfamine = sum(nonfamine)gen ins_famine = 1-sum_famine/sum_nonfaminedrop famine nonfamine sum_famine sum_nonfamine*extract data for county-level informationgenerate cr_info = [victims_cr!=.]generate grain_info = [grain_output!=.]generate school_info = !missing(speed_primary_density,speed_secondary_density )foreach var in yedu primary_graduate junior_graduate {replace `var' = . if treat !=0 | rural != 1 // only keep the baselinecollapse (mean) countyid pop1964 sdy_density han_ethn primary_graduate junior_graduate victims_cr cr_info grain_info school_info ins_famine, by(region1990)gen prov = floor(region1990/10000)save "$path1A\census_1990_county_char.dta", replace********************************************************************Table 2: Summary Statistics of the 1% Sample from the 1990 Census********************************************************************gen age = 1990 - year_birthoutsum yedu primary_graduate junior_graduate male han_ethn age if treat==0 & rural==1 using "$path4\Table2.txt", replaceoutsum yedu primary_graduate junior_graduate male han_ethn age if treat==0 & rural==0 using "$path4\Table2.txt", appendoutsum yedu primary_graduate junior_graduate male han_ethn age if treat==1 & rural==1 using "$path4\Table2.txt", appendoutsum yedu primary_graduate junior_graduate male han_ethn age if treat==1 & rural==0 using "$path4\Table2.txt", append**********************************************************************************************************************************************************************************************************************************************Table 3: The Effect of SDYs on the Educational Attainment of Rural Children******************************************************************************foreach var in yedu primary_graduate junior_graduate {if (`i'==1) reghdfe `var' c.sdy_density#c.treat male han_ethn if rural==1, absorb($var_abs_cohort) cluster(region1990)if (`i'==2) reghdfe `var' c.sdy_density#c.treat male han_ethn if rural==0, absorb($var_abs_cohort) cluster(region1990)summ `var' if e(sample)&treat==0if (("`var'"=="yedu")&(`i'==1)) outreg2 using "$path4\Table3.txt", replace se nonotes nocons noaster nolabel text addstat(Mean,`mean') keep(c.sdy_density#c.treat male han_ethn) sortvar(c.sdy_density#c.treat male han_ethn)if (("`var'"!="yedu")|(`i'!=1)) outreg2 using "$path4\Table3.txt", append se nonotes nocons noaster nolabel text addstat(Mean,`mean') keep(c.sdy_density#c.treat male han_ethn) sortvar(c.sdy_density#c.treat male han_ethn)drop rural // for the remaining analysis, we only use the rural samplegen treat_placebo = inrange(year_birth,1951,1955) if inrange(year_birth,1946,1955)reghdfe yedu c.sdy_density#c.treat_placebo male han_ethn, absorb($var_abs_cohort) cluster(region1990)outreg2 using "$path4\Table3.txt", append se nonotes nocons noaster nolabel text keep(c.sdy_density#c.treat c.sdy_density#c.treat_placebo male han_ethn) sortvar(c.sdy_density#c.treat c.sdy_density#c.treat_placebo male han_ethn)*****************************************************************************************Inputs for Figure 3: Effect of SDYs on the Educational Attainment of Different Cohorts*****************************************************************************************forvalues y = 1946/1969 {gen I`y' = sdy_density*[year_birth==`y']reghdfe yedu I1946-I1969 male han_ethn, absorb($var_abs_cohort2) cluster(region1990)outreg2 using "$path4\Figure3.txt", replace sideway noparen se nonotes nocons noaster nolabel text keep(I1946-I1969) sortvar(I1946-I1969)drop if inrange(year_birth,1941,1945) // these cohorts serve as the baseline in the above regression, and will not be used in the following analysis.****************************************Table 4: Heterogeneous Effect of SDYs****************************************if (`i'==1) reghdfe yedu c.sdy_density#c.treat male han_ethn if male==1 , absorb($var_abs_cohort) cluster(region1990)if (`i'==2) reghdfe yedu c.sdy_density#c.treat male han_ethn if male==0 , absorb($var_abs_cohort) cluster(region1990)if (`i'==3) reghdfe yedu c.sdy_density#c.treat male han_ethn if edu_base<5.5 , absorb($var_abs_cohort) cluster(region1990)if (`i'==4) reghdfe yedu c.sdy_density#c.treat male han_ethn if (edu_base>=5.5&edu_base<.), absorb($var_abs_cohort) cluster(region1990)if (`i'==5) reghdfe primary_graduate c.sdy_density#c.treat male han_ethn if edu_base<5.5 , absorb($var_abs_cohort) cluster(region1990)if (`i'==6) reghdfe junior_graduate c.sdy_density#c.treat male han_ethn if edu_base<5.5 , absorb($var_abs_cohort) cluster(region1990)if (`i'==7) reghdfe primary_graduate c.sdy_density#c.treat male han_ethn if (edu_base>=5.5&edu_base<.), absorb($var_abs_cohort) cluster(region1990)if (`i'==8) reghdfe junior_graduate c.sdy_density#c.treat male han_ethn if (edu_base>=5.5&edu_base<.), absorb($var_abs_cohort) cluster(region1990)summ yedu if e(sample)&treat==0summ primary_graduate if e(sample)&treat==0summ junior_graduate if e(sample)&treat==0if (`i'==1) outreg2 using "$path4\Table4.txt", replace se nonotes nocons noaster nolabel text addstat(Mean1,`mean1',Mean2,`mean2',Mean3,`mean3') keep(c.sdy_density#c.treat)if (`i'!=1) outreg2 using "$path4\Table4.txt", append se nonotes nocons noaster nolabel text addstat(Mean1,`mean1',Mean2,`mean2',Mean3,`mean3') keep(c.sdy_density#c.treat)********************************************************************************* Step 3: Contemporaneous Events and Other Outcome Variables ***********************************************************************************************************************************Table 6: Addressing Various Confounding Factors**************************************************replace grain_output = grain_output/pop1964replace victims_cr = victims_cr/pop1964gen treat_cr1 = inrange(year_birth,1954,1961)gen treat_cr2 = inrange(year_birth,1962,1968)gen famine_cohort1 = inrange(year_birth,1955,1958)gen famine_cohort2 = inrange(year_birth,1959,1961)*prepare for the interaction terms between school expansion program and SDYreghdfe yedu c.sdy_density#c.treat c.speed_primary_density#c.treat c.speed_secondary_density#c.treat male han_ethn, absorb($var_abs_cohort) cluster(region1990)summ speed_secondary_density if e(sample)==1gen DV_secondary_density = speed_secondary_density - r(mean)summ speed_primary_density if e(sample)==1gen DV_primary_density = speed_primary_density - r(mean)summ sdy_density if e(sample)==1gen DV_sdy_density = sdy_density - r(mean)local newvar1 "c.grain_output#c.treat"local newvar2 "c.speed_primary_density#c.treat c.speed_secondary_density#c.treat"local newvar3 "c.speed_primary_density#c.treat c.speed_secondary_density#c.treat c.DV_sdy_density#c.treat#c.DV_primary_density c.DV_sdy_density#c.treat#c.DV_secondary_density"local newvar4 "c.victims_cr#c.treat_cr1 c.victims_cr#c.treat_cr2"local newvar5 "c.ins_famine#c.famine_cohort1 c.ins_famine#c.famine_cohort2"local newvar6 "`newvar1' `newvar2' `newvar4' `newvar5'"local newvar_r c.grain_output#c.treat c.speed_primary_density#c.treat c.speed_secondary_density#c.treat c.DV_sdy_density#c.treat#c.DV_primary_density c.DV_sdy_density#c.treat#c.DV_secondary_density ///c.victims_cr#c.treat_cr1 c.victims_cr#c.treat_cr2 c.ins_famine#c.famine_cohort1 c.ins_famine#c.famine_cohort2if (`i'==1) local comm "replace"if (`i'!=1) local comm "append"reghdfe yedu c.sdy_density#c.treat `newvar`i'' male han_ethn, absorb($var_abs_cohort) cluster(region1990)outreg2 using "$path4\Table6_A.txt", `comm' se nonotes nocons noaster nolabel text keep(c.sdy_density#c.treat `newvar_r') sortvar(c.sdy_density#c.treat `newvar_r') replace sample = e(sample)reghdfe yedu c.sdy_density#c.treat male han_ethn if sample == 1, absorb($var_abs_cohort) cluster(region1990)outreg2 using "$path4\Table6_B.txt", `comm' se nonotes nocons noaster nolabel text keep(c.sdy_density#c.treat)***********************************************************************Table 8: The Lasting Effect of SDYs on Outcomes other than Education***********************************************************************gen senior_high = [yedu > 9] if yedu>=9 & yedu<. /*According to our definition of yedu, junior high graduates receive 9 years of education.Going beyond 9 years of education is equivalent to going beyond junior high education.*/gen occ_highskill = inlist(occisco,2,3) if !inlist(occisco,1,99)if (`i'==1) reghdfe senior_high c.sdy_density#c.treat male han_ethn, absorb($var_abs_cohort) cluster(region1990)if (`i'==2) reghdfe laborforce c.sdy_density#c.treat male han_ethn, absorb($var_abs_cohort) cluster(region1990)if (`i'==3) reghdfe laborforce c.sdy_density#c.treat yedu male han_ethn, absorb($var_abs_cohort) cluster(region1990)if (`i'==4) reghdfe occ_highskill c.sdy_density#c.treat male han_ethn, absorb($var_abs_cohort) cluster(region1990)if (`i'==5) reghdfe occ_highskill c.sdy_density#c.treat yedu male han_ethn, absorb($var_abs_cohort) cluster(region1990)if (`i'==6) reghdfe teacher c.sdy_density#c.treat male han_ethn, absorb($var_abs_cohort) cluster(region1990)if (`i'==7) reghdfe teacher c.sdy_density#c.treat yedu male han_ethn, absorb($var_abs_cohort) cluster(region1990)summ `e(depvar)' if e(sample)&treat==0if (`i'==1) outreg2 using "$path4\Table8.txt", replace se nonotes nocons noaster nolabel text addstat(Mean,`mean') keep(c.sdy_density#c.treat yedu) sortvar(c.sdy_density#c.treat yedu)if (`i'!=1) outreg2 using "$path4\Table8.txt", append se nonotes nocons noaster nolabel text addstat(Mean,`mean') keep(c.sdy_density#c.treat yedu) sortvar(c.sdy_density#c.treat yedu)
文中其他表格的code:
Table 3.txt Column (8) (The Effect of SDYs on the Educational Attainment of Rural Children)Table 5.txt (Effects of SDYs on the Supply of Local Teachers and Educational Fiscal Expenses, 1955--1977)Table 7.txt (The Effect of SDYs on Local People's Locus of Control)Table 8.txt Columns (8)--(10) (The Lasting Effect of SDYs on Outcomes other than Education)Figure 3.txt, census 1982/2010 ********************************************************************************* Step 1: Analysis using the 1982 Census *********************************************************************************global var_abs_cohort2 "region1982 prov#year_birth c.primary_base_older#year_birth c.junior_base_older#year_birth"use "$path1B\census_1982_clean.dta", clear*****************************************************************************************Inputs for Figure 3: Effect of SDYs on the Educational Attainment of Different Cohorts*****************************************************************************************forvalues y = 1946/1962 {gen I`y' = sdy_density*[year_birth==`y']reghdfe yedu I1946-I1962 male han_ethn, absorb($var_abs_cohort2) cluster(region1982)outreg2 using "$path4\Figure3.txt", append sideway noparen se nonotes nocons noaster nolabel text keep(I1946-I1962) sortvar(I1946-I1962)********************************************************************************* Step 2: Analysis using the 2000 Census *********************************************************************************global var_abs_cohort "region2000 prov#year_birth c.primary_base#year_birth c.junior_base#year_birth"global var_abs_cohort2 "region2000 prov#year_birth c.primary_base_older#year_birth c.junior_base_older#year_birth"use "$path1B\census_2000_clean.dta", clear******************************************************************************Table 3: The Effect of SDYs on the Educational Attainment of Rural Children******************************************************************************gen treat_placebo = inrange(year_birth,1975,1979) if inrange(year_birth,1970,1979)reghdfe yedu c.sdy_density#c.treat_placebo male han_ethn, absorb($var_abs_cohort) cluster(region2000)outreg2 using "$path4\Table3.txt", append se nonotes nocons noaster nolabel text keep(c.sdy_density#c.treat c.sdy_density#c.treat_placebo male han_ethn) sortvar(c.sdy_density#c.treat c.sdy_density#c.treat_placebo male han_ethn)*****************************************************************************************Inputs for Figure 3: Effect of SDYs on the Educational Attainment of Different Cohorts*****************************************************************************************forvalues y = 1946/1979 {gen I`y' = sdy_density*[year_birth==`y']reghdfe yedu I1946-I1979 male han_ethn, absorb($var_abs_cohort2) cluster(region2000)outreg2 using "$path4\Figure3.txt", append sideway noparen se nonotes nocons noaster nolabel text keep(I1946-I1979) sortvar(I1946-I1979)***********************************************************************Table 8: The Lasting Effect of SDYs on Outcomes other than Education***********************************************************************gen treat = inrange(year_birth,1956,1969) if inrange(year_birth,1946,1969)if (`i'==1) reghdfe age_marry1st c.sdy_density#c.treat male han_ethn, absorb($var_abs_cohort) cluster(region2000)if (`i'==2) reghdfe n_child c.sdy_density#c.treat male han_ethn, absorb($var_abs_cohort) cluster(region2000)summ `e(depvar)' if e(sample)&treat==0outreg2 using "$path4\Table8.txt", append se nonotes nocons noaster nolabel text addstat(Mean,`mean') keep(c.sdy_density#c.treat)********************************************************************************* Step 3: Analysis using the 2010 Census *********************************************************************************global var_abs_cohort "region2010 prov#year_birth c.primary_base#year_birth c.junior_base#year_birth"use "$path1B\census_2010_clean.dta", clear***********************************************************************Table 8: The Lasting Effect of SDYs on Outcomes other than Education***********************************************************************reghdfe yedu c.sdy_density#c.treat male han_ethn, absorb($var_abs_cohort) cluster(region2010)summ `e(depvar)' if e(sample)&treat==0outreg2 using "$path4\Table8.txt", append se nonotes nocons noaster nolabel text addstat(Mean,`mean') keep(c.sdy_density#c.treat) sortvar(c.sdy_density#c.treat)********************************************************************************* Step 4: Analysis using the 2010 CFPS *********************************************************************************global var_abs_cohort "region2010_h prov#year_birth c.primary_base#year_birth c.junior_base#year_birth"use "$path1B\CFPS_2010_clean.dta", cleargen treat = inrange(year_birth,1956,1969) if inrange(year_birth,1946,1969)foreach var of varlist LOC LOC_education LOC_talent LOC_effort LOC_hard_work LOC_intellect LOC_F_SES LOC_F_wealth LOC_F_connection LOC_luck LOC_connection {eststo: reghdfe `var' c.sdy_density#c.treat male han_ethn, vce(cluster region2010_h) absorb($var_abs_cohort)outreg2 [*] using "$path4\Table7_A.txt", se nonotes nocons noaster nolabel bdec(3) text replace keep(c.sdy_density#c.treat) sortvar(c.sdy_density#c.treat)foreach var of varlist LOC LOC_education LOC_talent LOC_effort LOC_hard_work LOC_intellect LOC_F_SES LOC_F_wealth LOC_F_connection LOC_luck LOC_connection {eststo: reghdfe `var' c.sdy_density#c.treat male han_ethn yedu, vce(cluster region2010_h) absorb($var_abs_cohort)outreg2 [*] using "$path4\Table7_B.txt", se nonotes nocons noaster nolabel bdec(3) text replace keep(c.sdy_density#c.treat yedu) sortvar(c.sdy_density#c.treat yedu)********************************************************************************* Step 5: Analysis using our county-by-year data *********************************************************************************use "$path1B\county_year_data.dta", clearkeep if inrange(year,1955,1977)gen postSDY = [year >= 1968] if inrange(year,1955,1977)foreach j in total state nonst {gen ratio_`i'_`j' = tch_`i'_`j'/pop1964gen fiscal_edu_pc = log(10000*fiscal_edu/pop1964)if (`i'==1) reghdfe ratio_pri_total c.sdy_density#c.postSDY, cluster(countyid) absorb(countyid year#prov)if (`i'==2) reghdfe ratio_pri_state c.sdy_density#c.postSDY, cluster(countyid) absorb(countyid year#prov)if (`i'==3) reghdfe ratio_pri_nonst c.sdy_density#c.postSDY, cluster(countyid) absorb(countyid year#prov)if (`i'==4) reghdfe ratio_sec_total c.sdy_density#c.postSDY, cluster(countyid) absorb(countyid year#prov)if (`i'==5) reghdfe ratio_sec_state c.sdy_density#c.postSDY, cluster(countyid) absorb(countyid year#prov)if (`i'==6) reghdfe ratio_sec_nonst c.sdy_density#c.postSDY, cluster(countyid) absorb(countyid year#prov)if (`i'==7) reghdfe fiscal_edu_pc c.sdy_density#c.postSDY, cluster(countyid) absorb(countyid year#prov)unique countyid if e(sample)if (`i'==1) outreg2 using "$path4\Table5.txt", se nonotes nocons noaster nolabel bdec(3) text replace addstat(Ncounty,`count') keep(c.sdy_density#c.postSDY) if (`i'!=1) outreg2 using "$path4\Table5.txt", se nonotes nocons noaster nolabel bdec(3) text append addstat(Ncounty,`count') keep(c.sdy_density#c.postSDY)
下面的code输出文章中的图：
Figure1.pdf (Number of SDYs by Resettlement, 1962--1979)Figure3.pdf (Effect of SDYs on the Educational Attainment of Different Cohorts)***************************************************************************Figure 1: Number of SDYs by Resettlement, 1962--1979 (Source: Gu (2009))***************************************************************************input str9 year total rural_village collective_farm state_farm1962-1966 129.28 87.06 0 42.221967-1968 199.68 165.96 0 33.721969 267.38 220.44 0 46.941974 172.48 119.19 34.63 18.661975 236.86 163.45 49.68 23.731976 188.03 122.86 41.51 23.661977 171.68 113.79 41.9 15.991978 48.09 26.04 18.92 3.131979 24.77 7.32 16.44 1.01foreach var in total rural_village collective_farm state_farm {gen v_temp = rural_village + collective_farmencode year, generate(period)twoway bar rural_village period, barw(0.6) base(0) color(gs2) ///|| rbar v_temp rural_village period, barw(0.6) color(gs12) ///|| rbar total v_temp period, barw(0.6) color(gs7) ///||, graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(0(0.5)3, angle(0) format(%12.1f)) ytitle("Number of SDYs (Million)", margin(medium)) /// xlabel(1(1)13, noticks valuelabel angle(90)) xtitle("Year") /// legend(label(1 "Rural Villages") label(2 "Collective Farms") label(3 "State Farms") ring(0) pos(2) colgap(*0.5) ) graph export "$path4\Figure1.pdf",replace******************************************************************************Figure 3: Effect of SDYs on the Educational Attainment of Different Cohorts******************************************************************************insheet using "$path4\Figure3.txt", cleargen year = substr(v1,2,4)rename (v2 v3 v4 v5 v6 v7)(coef1990 se1990 coef1982 se1982 coef2000 se2000)reshape long coef se, i(year) j(data)gen y_overlap = min(max(year-1955,0),max(1970-year,0),6)twoway line lb year if data==1982, sort lpattern(dash) lcolor(gs8) yaxis(1) ///|| line ub year if data==1982, sort lpattern(dash) lcolor(gs8) ///|| line coef year if data==1982, lwidth(thick) lcolor(black) yaxis(1) ///|| line y_overlap year if data==1982, sort lpattern(dash_dot) lwidth(thick) lcolor(gs8) yaxis(2) ///||, graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(-4(2)8, labsize(small) angle(0) format(%12.0f) axis(1)) ytitle("Coefficients", size(small) axis(1)) /// ylabel(0(2)6, labsize(small) angle(0) format(%12.0f) axis(2)) ytick(-6 0(1)6 12,axis(2)) ytitle("Years of Overlap", size(small) axis(2)) /// xlabel(1945(5)1980, labsize(small)) xtick(1945(5)1980) xtitle("Birth Cohort", size(small)) /// xline(1955 1970, lpattern(solid) lwidth(thin) lcolor(black)) /// title("Panel A - Census 1982", size(small) margin(small)) /// yline(0, lpattern(solid) lwidth(thin) lcolor(black)) legend(off) fxsize(70) fysize(60)twoway line lb year if data==1990, lpattern(dash) lcolor(gs8) yaxis(1) ///|| line ub year if data==1990, lpattern(dash) lcolor(gs8) ///|| line coef year if data==1990, lwidth(thick) lcolor(black) yaxis(1) ///|| line y_overlap year if data==1990, lpattern(dash_dot) lwidth(thick) lcolor(gs8) yaxis(2) ///||, graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(-4(2)8, labsize(small) angle(0) format(%12.0f) axis(1)) ytitle("Coefficients", size(small) axis(1)) /// ylabel(0(2)6, labsize(small) angle(0) format(%12.0f) axis(2)) ytick(-6 0(1)6 12,axis(2)) ytitle("Years of Overlap", size(small) axis(2)) /// xlabel(1945(5)1980, labsize(small)) xtick(1945(5)1980) xtitle("Birth Cohort", size(small)) /// xline(1955 1970, lpattern(solid) lwidth(thin) lcolor(black)) /// title("Panel B - Census 1990", size(small) margin(small)) /// yline(0, lpattern(solid) lwidth(thin) lcolor(black)) legend(off) fxsize(70) fysize(60)twoway line lb year if data==2000, lpattern(dash) lcolor(gs8) yaxis(1) ///|| line ub year if data==2000, lpattern(dash) lcolor(gs8) ///|| line coef year if data==2000, lwidth(thick) lcolor(black) yaxis(1) ///|| line y_overlap year if data==2000, lpattern(dash_dot) lwidth(thick) lcolor(gs8) yaxis(2) ///||, graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(-3(1)6, labsize(small) angle(0) format(%12.0f) axis(1)) ytitle("Coefficients", size(small) axis(1)) /// ylabel(0(2)6, labsize(small) angle(0) format(%12.0f) axis(2)) ytick(-6 0(1)6 12,axis(2)) ytitle("Years of Overlap", size(small) axis(2)) /// xlabel(1945(5)1980, labsize(small)) xtick(1945(5)1980) xtitle("Birth Cohort", size(small)) /// legend(order(3 1 4)label(3 "Coefficient") label(1 "95% CI") label(4 "Overlapped Years in""Primary Schools") col(2) size(small) margin(tiny)) /// xline(1955 1970, lpattern(solid) lwidth(thin) lcolor(black)) /// title("Panel C - Census 2000", size(small) margin(small)) /// yline(0, lpattern(solid) lwidth(thin) lcolor(black)) fxsize(65) fysize(80)twoway || connected coef year if data==1982, lwidth(medthick) msymbol(triangle) color(black) ///|| line coef year if data==1990, lwidth(medthick) color(gs6) ///|| connected coef year if data==2000, lwidth(medthick) msymbol(square) color(gs12) ///||, graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(-2(1)5, labsize(small) angle(0) format(%12.0f)) ytitle("Coefficients", size(small)) /// xlabel(1945(5)1980, labsize(small)) xtick(1945(5)1980) xtitle("Birth Cohort", size(small)) /// legend(label(1 "Census 1982") label(2 "Census 1990") label(3 "Census 2000") col(2) size(small)) /// xline(1955 1970, lpattern(solid) lwidth(thin) lcolor(black)) /// title("Panel D - Three Censuses in One Graph", size(small) margin(small)) /// yline(0, lpattern(solid) lwidth(thin) lcolor(black)) fxsize(70) fysize(80)graph combine a.gph b.gph c.gph d.gph, graphregion(fcolor(gs16) lcolor(gs16))graph export "$path4\Figure3.pdf",replaceerase "$path4\Figure3.txt"附录的表格code：
****************************************************************************************************************************************************************use "$path1B\census_1990_clean.dta", clearglobal var_abs_cohort "region1990 prov#year_birth c.primary_base#year_birth c.junior_base#year_birth"keep if inrange(year_birth,1946,1969)gen treat = inrange(year_birth,1956,1969) if inrange(year_birth,1946,1969)************************************************************ generate base education level for both rural and urban ************************************************************gen edu_temp_urban = yedu if treat == 0 & rural == 0 gen prefec = floor(region1990/100)bysort region1990: egen edu_base_urban1 = mean(edu_temp_urban)bysort prefec : egen edu_base_urban2 = mean(edu_temp_urban)bysort prov : egen edu_base_urban3 = mean(edu_temp_urban)*************************************************** Table A1: Knowledge Gap and the Effect of SDYs ***************************************************gen edu_base_diff`i' = edu_base_urban`i' - edu_basesumm edu_base_diff`i' if !missing(yedu,sdy_density,edu_base_diff`i',treat) & rural==1gen DV_edu_base_diff`i' = edu_base_diff`i' - r(mean)summ sdy_density if !missing(yedu,sdy_density,edu_base_diff`i',treat) & rural==1gen DV_sdy_density = sdy_density - r(mean) reghdfe yedu c.sdy_density#c.treat c.treat#c.edu_base_diff`i' c.DV_sdy_density#c.treat#c.DV_edu_base_diff`i' male han_ethn if rural==1, absorb($var_abs_cohort) cluster(region1990)if (`i'==1) outreg2 using "$path4\TableA1.txt", replace se nonotes nocons noaster nolabel text keep(c.sdy_density#c.treat c.DV_sdy_density#c.treat#c.DV_edu_base_diff`i')if (`i'!=1) outreg2 using "$path4\TableA1.txt", append se nonotes nocons noaster nolabel text keep(c.sdy_density#c.treat c.DV_sdy_density#c.treat#c.DV_edu_base_diff`i')drop DV_edu_base_diff`i' DV_sdy_densitydrop edu_base_diff* edu_base_urban********************************************************* Table A3: The Effect of SDYs on Occupational Choice ********************************************************gen O`i' = [occisco==`i'] if occisco!=99reghdfe O`i' c.sdy_density#c.treat male han_ethn if rural==1, absorb($var_abs_cohort) cluster(region1990)summ `e(depvar)' if e(sample)&treat==0if (`i'==1) outreg2 using "$path4\TableA3.txt", replace se nonotes nocons noaster nolabel text addstat(Mean,`mean') keep(c.sdy_density#c.treat)if (`i'!=1) outreg2 using "$path4\TableA3.txt", append se nonotes nocons noaster nolabel text addstat(Mean,`mean') keep(c.sdy_density#c.treat)************************************************************************************************************************************************************************************************************************************** Prepare the information availability of county-by-year level data **********************************************************************use "$path1B\county_year_data.dta", clearkeep if inrange(year,1955,1977)bysort countyid: egen count1 = count(tch_sec_total)bysort countyid: egen count2 = count(tch_pri_total)bysort countyid: egen count3 = count(fiscal_edu)gen share_nonmissing_teacher = (count1 + count2)/46gen share_nonmissing_fiscal = (count3)/23keep countyid share_nonmissing_teacher share_nonmissing_fiscalsave "$path1A\teacher_fiscal_info.dta", replace****************************************** Table B1: Count of Number of Counties ******************************************use "$path1B\county_data.dta", clearmerge 1:1 countyid using "$path1A\teacher_fiscal_info.dta", nogenerate keep(1 3)merge m:1 countyid using "$path1A\rural_school_expansion.dta", nogenerate keep(1 3)replace share_nonmissing_teacher = 0 if share_nonmissing_teacher == .replace share_nonmissing_fiscal = 0 if share_nonmissing_fiscal == .scalar r1 = r(unique) // number in Panel A, row 1gen prov = floor(region1990/10000)unique region1990 if !inlist(prov,11,12,31)scalar r2 = r(unique) // number in Panel A, row 2unique region1990 if !inlist(prov,11,12,31) & district!=1scalar r3 = r(unique) // number in Panel A, row 3unique region1990 if !inlist(prov,11,12,31) & district!=1 & sdy!=.scalar r4 = r(unique) // number in Panel A, row 4unique region1990 if !inlist(prov,11,12,31) & district!=1 & sdy!=. & pop1964!=.scalar r5 = r(unique) // number in Panel A, row 5*Panel B is conditional on "core counties"keep if !inlist(prov,11,12,31) & district!=1 & sdy!=. & pop1964!=.unique region1990 if grain_output !=.scalar r9 = r(unique) // number in Panel B, row 3unique region1990 if !missing(secondary_speed,primary_speed)scalar r12 = r(unique) // number in Panel B, row 6unique region1990 if !missing(victims_cr)scalar r13 = r(unique) // number in Panel B, row 7*For the following variables, they don't have to show up in the 1990 census.use "$path1B\county_data.dta", clearmerge 1:1 countyid using "$path1A\teacher_fiscal_info.dta", nogenerate keep(1 3)merge m:1 countyid using "$path1A\rural_school_expansion.dta", nogenerate keep(1 3)replace share_nonmissing_teacher = 0 if share_nonmissing_teacher == .replace share_nonmissing_fiscal = 0 if share_nonmissing_fiscal == .gen prov = floor(region1990/10000)keep if !inlist(prov,11,12,31) & district!=1 & sdy!=. & pop1964!=.unique countyid if share_nonmissing_teacher > 0scalar r7 = r(unique) // number in Panel B, row 1summ share_nonmissing_teacher if share_nonmissing_teacher > 0 scalar r8 = r(mean) // number in Panel B, row 2unique countyid if share_nonmissing_fiscal > 0scalar r10 = r(unique) // number in Panel B, row 4summ share_nonmissing_fiscal if share_nonmissing_fiscal > 0 scalar r11 = r(mean) // number in Panel B, row 5replace num = r`i' in `i'outsheet using "$path4\TableB1.txt", replace*************************************************************************************************** Table B2: Correlation between County-level Information Availability and County Characteristics ***************************************************************************************************use "$path1A\census_1990_county_char.dta", clearmerge 1:1 countyid using "$path1A\teacher_fiscal_info.dta", keep(1 3) nogenerategen minority = 1 - han_ethnreplace victims_cr = victims_cr/pop1964replace share_nonmissing_teacher = 0 if share_nonmissing_teacher == .replace share_nonmissing_fiscal = 0 if share_nonmissing_fiscal == .eststo: reghdfe sdy_density primary_graduate junior_graduate minority ins_famine, vce(robust) absorb(prov)eststo: reghdfe sdy_density victims_cr primary_graduate junior_graduate minority ins_famine, vce(robust) absorb(prov)eststo: reghdfe share_nonmissing_teacher sdy_density primary_graduate junior_graduate minority ins_famine, vce(robust) absorb(prov)eststo: reghdfe grain_info sdy_density primary_graduate junior_graduate minority ins_famine, vce(robust) absorb(prov)eststo: reghdfe share_nonmissing_fiscal sdy_density primary_graduate junior_graduate minority ins_famine, vce(robust) absorb(prov)eststo: reghdfe school_info sdy_density primary_graduate junior_graduate minority ins_famine, vce(robust) absorb(prov)eststo: reghdfe cr_info sdy_density primary_graduate junior_graduate minority ins_famine, vce(robust) absorb(prov)global var_order "sdy_density victims_cr ins_famine primary_graduate junior_graduate minority"outreg2 [*] using "$path4\TableB2.txt", replace se nonotes nocons noaster nolabel text keep($var_order) sortvar($var_order)***************************************************************************************************************************************************************************************************************************************** Table C1: Comparing the Number of Received SDYs from County-aggregate** with that from National Report in Each Province, Column (1) ** Note: Column (2) comes from Gu (2009) *************************************************************************use "$path1B\county_data.dta", cleargen prov = floor(region1990/10000)drop if inlist(prov,11,12,31,54) // Drop Beijing, Tianjin, Shanghai, and Tibetreplace prov = 44 if prov == 46 // Hainan is part of Guangdong during the movementreplace prov = 51 if prov == 50 // Chongqing is part of Sichuan during the movementcollapse (sum) sdy, by(prov)outsheet using "$path4\TableC1_column1.txt", replace****************************************************************************************************************************************************************use "$path1B\census_1990_clean.dta", clearglobal var_abs_cohort "region1990 prov#year_birth c.primary_base#year_birth c.junior_base#year_birth"gen treat = inrange(year_birth,1956,1969) if inrange(year_birth,1946,1969)gen treat_alt = inrange(year_birth,1956,1969) if inrange(year_birth,1946,1969)&!inrange(year,1953,1955)gen treat_junior = inrange(year_birth,1953,1969) if inrange(year_birth,1943,1969)keep if inrange(year_birth,1943,1969) ************************************************************* Table D1: Robustness Check with Different Specifications **************************************************************alternative densities of SDYbysort region1990: egen cohort_size = sum(treat)gen sdy_density_alt = sdy_density*pop1964/(100*cohort_size)*alternative exposure to SDYgen primary_overlap = min(max(year-1955,0),max(1970-year,0),6) if inrange(year_birth,1946,1969)gen junior_overlap = min(max(year-1952,0),max(1970-year,0),9) if inrange(year_birth,1943,1969)global var_order "c.sdy_density#c.treat c.sdy_density_alt#c.treat c.sdy_density#c.primary_overlap c.sdy_density#c.treat_alt c.sdy_density#c.treat_junior c.sdy_density#c.junior_overlap"if (`i'>=1&`i'<=3) local dep_var "c.sdy_density#c.treat"if (`i'==4) local dep_var "c.sdy_density_alt#c.treat"if (`i'==5) local dep_var "c.sdy_density#c.primary_overlap"if (`i'==6) local dep_var "c.sdy_density#c.treat_alt"if (`i'==7) local dep_var "c.sdy_density#c.treat_junior"if (`i'==8) local dep_var "c.sdy_density#c.junior_overlap"if (`i'==1) local cond "& year_birth<=1966"if (`i'==2) local cond "& year_birth<=1963"if (`i'==3) local cond "& year_birth<=1960"reghdfe yedu `dep_var' male han_ethn if rural==1 `cond', absorb($var_abs_cohort) cluster(region1990)if (`i'==1) outreg2 using "$path4\TableD1_A.txt", se nonotes nocons noaster nolabel text replace keep($var_order) sortvar($var_order)if (`i'!=1) outreg2 using "$path4\TableD1_A.txt", se nonotes nocons noaster nolabel text append keep($var_order) sortvar($var_order)replace temp_var = `dep_var' // to make the table easier to read, Panel B reports the corresponding coefficients to Panel Areghdfe yedu temp_var male han_ethn if rural==0 `cond', absorb($var_abs_cohort) cluster(region1990)if (`i'==1) outreg2 using "$path4\TableD1_B.txt", se nonotes nocons noaster nolabel text replace keep(temp_var)if (`i'!=1) outreg2 using "$path4\TableD1_B.txt", se nonotes nocons noaster nolabel text append keep(temp_var)************************************** Table D2: Other Robustness Checks ***************************************drop nine Third-Frontier provinceseststo: reghdfe yedu c.sdy_density#c.treat male han_ethn if rural==1 & !inlist(prov,51,52,61,62,42,43,53,64,65), absorb($var_abs_cohort) cluster(region1990)eststo: reghdfe yedu c.sdy_density#c.treat male han_ethn if rural==0 & !inlist(prov,51,52,61,62,42,43,53,64,65), absorb($var_abs_cohort) cluster(region1990)*drop five provinces that does not match well between local gazettes and national reports eststo: reghdfe yedu c.sdy_density#c.treat male han_ethn if rural==1 & !inlist(prov,14,23,53,64,65), absorb($var_abs_cohort) cluster(region1990)eststo: reghdfe yedu c.sdy_density#c.treat male han_ethn if rural==0 & !inlist(prov,14,23,53,64,65), absorb($var_abs_cohort) cluster(region1990)*impose stronger assumptions on migration historyeststo: reghdfe yedu c.sdy_density#c.treat male han_ethn if rural==1 & local_1985==1, absorb($var_abs_cohort) cluster(region1990)eststo: reghdfe yedu c.sdy_density#c.treat male han_ethn if rural==0 & local_1985==1, absorb($var_abs_cohort) cluster(region1990)*drop sample whose education are eligible for hukou transitioneststo: reghdfe yedu c.sdy_density#c.treat male han_ethn if rural==1 & hukou_transit==0, absorb($var_abs_cohort) cluster(region1990)eststo: reghdfe yedu c.sdy_density#c.treat male han_ethn if rural==0 & hukou_transit==0, absorb($var_abs_cohort) cluster(region1990)*drop counties that SDY numbers end with zerogen last_digit = mod(sdy,10)eststo: reghdfe yedu c.sdy_density#c.treat male han_ethn if rural==1 & last_digit!=0, absorb($var_abs_cohort) cluster(region1990)eststo: reghdfe yedu c.sdy_density#c.treat male han_ethn if rural==0 & last_digit!=0, absorb($var_abs_cohort) cluster(region1990)outreg2 [*] using "$path4\TableD2.txt", se nonotes nocons noaster nolabel text replace keep(c.sdy_density#c.treat)erase "$path1A\census_1990_county_char.dta"erase "$path1A\rural_school_expansion.dta"erase "$path1A\teacher_fiscal_info.dta"Figure A1.pdf (Trends of Real Educational Expenditures in Local Gazetteers)Figure A2.pdf (The Process of China's Secondary Education Expansion since the Late 1960s)Figure C1.pdf (Benford's Law and Data Quality on SDYs)Figure C2.pdf (Number of SDYs Estimated from CFPS 2010)Figure E1.pdf (Estimating the Effect of SDYs using the Synthetic Control Method)********************************************************************************* Figures in Appendix A **********************************************************************************************************************************************************Figure A1: Trends of Real Educational Expenditures in Local Gazetteers*************************************************************************use "$path1B\county_year_data.dta", clearkeep if inrange(year,1950,1990)rename fiscal_edu fiscal_edu_countycollapse (mean) fiscal_edu_county, by(year)merge 1:1 year using "$path1B\NBS_data.dta", nogenerate keepusing(fiscal_edu price_deflator)replace fiscal_edu_national = fiscal_edu_national/price_deflatorreplace fiscal_edu_county = fiscal_edu_county/price_deflatortwoway line fiscal_edu_national year, yaxis(1) lcolor(black) lpattern(solid) ///|| line fiscal_edu_county year, yaxis(2) lcolor(gs8) lpattern(dash) ///||, graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(0(50)250, angle(0) format(%12.0f) axis(1)) ytitle("National Fiscal Educational Expenditures""from NBS (100 million RMB)", margin(medium) axis(1)) /// ylabel(0(100)500, angle(0) format(%12.0f) axis(2)) ytitle("Educational Expenditures Per County""from Local Gazeteers (10,000 RMB)", margin(medium) axis(2)) /// xlabel(1950(5)1990) xtick(1950(5)1990) xtitle("Year") /// legend(label(1 "National Fiscal Educational""Expenditures from NBS") label(2 "Educational Expenditures Per County""from Local Gazeteers") col(1) size(medsmall))graph export "$path4\FigureA1.pdf",replace***************************************************************************************Figure A2: The Process of China's Secondary Education Expansion since the Late 1960s***************************************************************************************use "$path1B\county_year_data.dta", clearcollapse (mean) school_primary school_secondary, by(year)twoway line school_primary year, yaxis(1) lcolor(black) lpattern(solid) ///|| line school_secondary year, yaxis(2) lcolor(gs8) lpattern(dash) ///||, graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(0(150)600, angle(0) format(%12.0f) axis(1)) ytitle("# Primary Schools per County", margin(medium) axis(1)) /// ylabel(0(20)80, angle(0) format(%12.0f) axis(2)) ytitle("# Secondary Schools per County", margin(medium) axis(2)) /// xlabel(1950(5)1990) xtick(1950(5)1990) xtitle("Year") title("Panel A - Summary Statistics from Local Gazeteers",size(medium) margin(medium)) /// legend(label(1 "# Primary Schools per County") label(2 "# Secondary Schools per County") col(1) size(medsmall)) use "$path1B\NBS_data.dta", cleartwoway line primary_stu year, yaxis(1) lcolor(black) lpattern(solid) ///|| line secondary_stu year, yaxis(2) lcolor(gs8) lpattern(dash) ///||, graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(0(400)1600, angle(0) format(%12.0f) axis(1)) ytitle("# Primary Students per 10,000", margin(medium) axis(1)) /// ylabel(0(200)800, angle(0) format(%12.0f) axis(2)) ytitle("# Secondary Students per 10,000", margin(medium) axis(2)) /// xlabel(1950(5)1990) xtick(1950(5)1990) xtitle("Year") title("Panel B - National-level Statistics",size(medium) margin(medium)) /// legend(label(1 "# Primary Students per 10,000") label(2 "# Secondary Students per 10,000") col(1) size(medsmall))graph combine a.gph b.gph, rows(2) graphregion(fcolor(gs16) lcolor(gs16)) xsize(13.5) ysize(20)graph export "$path4\FigureA2.pdf",replace********************************************************************************* Figures in Appendix C *************************************************************************************************************************************Figure C1: Benford's Law and Data Quality on SDYs****************************************************use "$path1B\county_data.dta", cleargen prov = floor(region1990/10000)keep if !inlist(prov,11,12,31) & district!=1 & sdy_density!=. // keep the sample corresponding to our main analysis. firstdigit sdy, percent n chi-sq. P-value digit observed expected------------------------------------------------------------sdy 1773 6.81 0.5575 1 28.54 30.10input str9 digit data benfordend // the input comes from the results of firstdight, as shown abovegraph bar benford data, over(digit) bargap(0) bar(1,color(gs0)) bar(2,color(gs12)) ///graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(0(5)30, angle(0) format(%12.0f)) ytitle("Percentage Points", margin(medium)) /// b1title("First Digit", margin(medium)) /// legend(label(1 "Data on SDYs") label(2 "Benford's Law") ring(0) pos(2) colgap(*0.5)) graph export "$path4\FigureC1.pdf",replace *****************************************************Figure C2: Number of SDYs Estimated from CFPS 2010*****************************************************/*Note: Plotting this graph requires the original CFPS 2010 data.We directly provide the output numbers here. Those numbers can bereplicated with the following codes. */use "$path2B\cfps2010adult_201906",clearrename qg101_a_1 sdy_startrename qa1y_best year_birthkeep if qg1_s_1r==1|qg1_s_2r==1 // keep SDY samplekeep if inrange(sdy_start,1962,1979)recode sdy_start (1962/1966=1)(1967/1968=2), gen(period)replace period = sdy_start-1966 if inrange(sdy_start,1969,1979)collapse (sum) rswt_nat, by(period)gen cfps_impute = rswt_nat/10000input str9 year sdy_national cfps_imputereplace sdy_national= sdy_national/100replace cfps_impute = cfps_impute/100encode year, generate(period)graph bar sdy_national cfps_impute, over(period,label(angle(90))) bargap(0) bar(1,color(gs0)) bar(2,color(gs12)) ///graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(0(0.5)3, angle(0) format(%12.1f)) ytitle("Number of SDYs (Million)", margin(medium)) /// legend(label(1 "National Reports") label(2 "CFPS Estimates") ring(0) pos(2) colgap(*0.5) ) graph export "$path4\FigureC2.pdf",replace ********************************************************************************* Figures in Appendix E ************************************************************************************************************************** Synthetic Control, Step 1: ** prepare county-level characteristics *****************************************use "$path1B\census_1990_clean.dta", cleargen minority = 1 - han_ethngen famine = inrange(year_birth,1959,1961)gen nonfamine = inrange(year_birth,1955,1957)replace minority = . if !inrange(year_birth,1946,1955)replace primary_graduate = . if !inrange(year_birth,1946,1955)replace junior_graduate = . if !inrange(year_birth,1946,1955)collapse (mean) minority primary_graduate junior_graduate (sum) famine nonfamine, by(region1990)gen ins_famine = 1 - famine/nonfaminesave "$path1A\census_1990_county_SC.dta", replace***************************************************************** Synthetic Control, Step 2: ** pick up counties with sufficient observations in each cohort *****************************************************************use "$path1B\census_1990_clean.dta", clearkeep if inrange(year_birth,1946,1969) collapse (count) N = yedu, by(year_birth region1990)bysort region1990: gen balance = _Nsave "$path1A\census_1990_county_list.dta", replace****************************************************** Synthetic Control, Step 3: ** prepare county-by-cohort data for the SC analysis ******************************************************use "$path1B\county_data.dta", cleardrop if region1990 == . |countyid == .drop region1982 region2000 region2010*SC requires counties to have complete informationmerge 1:1 region1990 using "$path1A\census_1990_county_list.dta", keep(3) nogeneratemerge 1:1 region1990 using "$path1A\census_1990_county_SC.dta", keep(3) nogeneratereplace grain_output = grain_output/pop1964keep region1990 sdy_density grain_output minority ins_famine urbanratio64 primary_graduate junior_graduatekeep if !missing(sdy_density,grain_output,minority,ins_famine,urbanratio64)gen treat = [_n > (N+1)/2] use "$path1B\census_1990_clean.dta", clearkeep if inrange(year_birth,1946,1969) collapse (mean) yedu, by(year_birth region1990)merge m:1 region1990 using `temp', keep(3) nogeneratetsset region1990 year_birthsave "$path1A\synth_analysis.dta", replace****************************************************** Synthetic Control, Step 4: ** extended Abadie synthetic control (SC) method ******************************************************use "$path1A\synth_analysis.dta", cleargen D = [year_birth>=1956]*treatsynth_runner yedu yedu(1946(1)1955) primary_graduate(1955) junior_graduate(1955) grain_output(1955) urbanratio64(1955) minority(1955) ins_famine(1955), d(D) gen_var parallelsave "$path1A\synth_results.dta", replacereshape long Pc, i(I) j(lead)use "$path1A\synth_results.dta", clearmerge m:1 lead using `temp', nogeneratesave "$path1A\synth_results.dta", replace******************************************************************************Figure E1: Estimating the Effect of SDYs using the Synthetic Control Method******************************************************************************use "$path1A\synth_results.dta", clearcollapse (mean) p_vals yedu yedu_synth, by(year_birth)gen effect = yedu - yedu_synthtwoway line yedu year_birth, lcolor(black) || line yedu_synth year_birth, lpattern(dash) lcolor(black) ///||, graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(5(0.5)8, angle(0) format(%12.1f)) ytitle("Average Years of Education", margin(medium)) /// xlabel(1945(5)1970) xtick(1945(5)1970) xtitle("Birth Cohort") /// legend(label(1 "Treated (Counties with Higher""Density of SDY)") label(2 "Synthetic Control from Counties""with Lower Density of SDY") col(1) size(small) ring(0) pos(4) colgap(*0.5)) /// xline(1955, lpattern(solid) lwidth(thin) lcolor(black)) /// title("Treatment versus Control",size(medium) margin(medium))twoway line effect year_birth, yaxis(1) lcolor(black) || scatter p_vals year_birth, yaxis(2) msymbol(square) mcolor(black) ///||, graphregion(fcolor(gs16) lcolor(gs16)) plotregion(lcolor(gs16) margin(zero)) /// ylabel(-0.1(0.1)0.4, angle(0) format(%12.1f) axis(1)) ytitle("Treatment Effect", margin(medium) axis(1)) /// ylabel(0(0.2)1, angle(0) format(%12.1f) axis(2)) ytick(-0.25 0(0.2)1,axis(2)) ytitle("p-values", margin(medium) axis(2)) /// xlabel(1945(5)1970) xtick(1945(5)1970) xtitle("Birth Cohort") /// legend(label(1 "Treatment Effect") label(2 "p-values") col(1) size(small) ring(0) pos(3) colgap(*0.5)) /// xline(1955, lpattern(solid) lwidth(thin) lcolor(black)) /// title("Treatment Effect and P-values",size(medium) margin(medium))graph combine a.gph b.gph, rows(1) graphregion(fcolor(gs16) lcolor(gs16)) xsize(18) ysize(8)graph export "$path4\FigureE1.pdf",replaceerase "$path1A\census_1990_county_list.dta"erase "$path1A\census_1990_county_SC.dta"erase "$path1A\synth_analysis.dta"erase "$path1A\synth_results.dta"
关于相关计量方法视频课程，文章，数据和代码，参看 1.面板数据方法免费课程, 文章, 数据和代码全在这里, 优秀学人好好收藏学习！2.双重差分DID方法免费课程, 文章, 数据和代码全在这里, 优秀学人必须收藏学习！3.工具变量IV估计免费课程, 文章, 数据和代码全在这里, 不学习可不要后悔！4.各种匹配方法免费课程, 文章, 数据和代码全在这里, 掌握匹配方法不是梦！5.断点回归RD和合成控制法SCM免费课程, 文章, 数据和代码全在这里, 有必要认真研究学习！6.空间计量免费课程, 文章, 数据和代码全在这里, 空间相关学者注意查收！
下面这些短链接文章属于合集，可以收藏起来阅读，不然以后都找不到了。